Robotic AI Systems for Fake News Detection in IoT-Connected Social Media Platforms Using Sensor-Driven Cross-Verification

Sakera Begum; Md Ismail Jobi Ullah; Mohammad Kabir Hussain; Sanjida Alam  Eshra; Amjad Hossain; Md Arifur Rahaman; Md Shadman Soumik; Md Shohel Rana  Palleb; Mrinmoy Sarkar; Md Mustafizur Rahman

doi:10.63332/joph.v5i11.3688

Authors

Sakera Begum Washington University of Science and Technology, 2900 Eisenhower Ave, Alexandria, VA 22314
Md Ismail Jobi Ullah Washington University of Science and Technology, 2900 Eisenhower Ave, Alexandria, VA 22314
Mohammad Kabir Hussain Washington University of Science and Technology, 2900 Eisenhower Ave, Alexandria, VA 22314
Sanjida Alam Eshra Trine University, USA
Amjad Hossain Department: Business Analytics, School of Business University: Mercy University, USA
Md Arifur Rahaman Degree & Institution: MS in Project Management, St. Francis College, Brooklyn, NY, USA
Md Shadman Soumik Student, Master of Science in Information Technology Washington University of Science and Technology
Md Shohel Rana Palleb Student (Ex.), Department of Agricultural Extension Education Bangladesh Agricultural University (BAU)
Mrinmoy Sarkar Department: Master’s in Information Technology (MSIT) Washington University of Science and Technology
Md Mustafizur Rahman Master’s in Computer Science University: Mercy University

DOI:

https://doi.org/10.63332/joph.v5i11.3688

Keywords:

Fake news detection, RoBERTa, IoT sensors, Cross-verification, Hybrid AI

Abstract

Much of the fake news and misinformation peddling can be attributed to this quick development in Internet of Things, and the web related social media platforms. In order to enable the detection of actual fake news, our present research suggests a robotic AI model through text and sensor data. The combination of the holistic model that is suggested consists of sensor-based cross-checking (confidence, location, time synchronization and anomaly detection) and RoBERTa transformer models to interpolate textual contents. Baselines were also used to compare model with a PolitiFact, LIAR and FakenewsNet datasets baselines of text only and sensor data only. Experimental results have demonstrated that the hybrid strategy has shown improved performance with all results, with much more accurate and reliable detection with the consideration of physical context. Findings indicate the potential of sensor-enhanced AI-based systems to reduce the risk of misinformation with regard to IoT-connecting ecosystems, which may inform the course of action with regard to the development of reliable, smart and context-sensitive digital media surveillance systems.